Natural gas and other gases are liquefied for storage and transport. Liquefaction reduces the volume of the gas and is typically carried out by chilling the gas through indirect heat exchange in one or more refrigeration cycles. The refrigeration cycles are costly because of the complexity of the equipment and the performance efficiency of the cycle. There is a need, therefore, for gas cooling and/or liquefaction systems that are less complex, more efficient, and less expensive to operate.
Liquefying natural gas, which is primarily methane, typically requires cooling the gas stream to approximately −160° C. to −170° C. and then letting down the pressure to approximately atmospheric. Typical temperature-enthalpy curves for liquefying gaseous methane, such as shown in FIG. 1 (methane at 60 bar pressure, methane at 35 bar pressure, and a methane/ethane mixture at 35 bar pressure), have three regions along an S-shaped curve. As the gas is cooled, at temperatures above about −75° C. the gas is de-superheating; and at temperatures below about −90° C. the liquid is subcooling. Between these temperatures, a relatively flat region is observed—in which the gas is condensing into liquid. In the 60 bar methane curve, because the gas is above the critical pressure, only one phase is present above the critical temperature, but its specific heat is large near the critical temperature; below the critical temperature the cooling curve is similar to the lower pressure (35 bar) curves. The 35 bar curve for 95% methane/5% ethane shows the effect of impurities, which round off the dew and bubble points.
Refrigeration processes supply the requisite cooling for liquefying natural gas, and the most efficient of these have heating curves that closely approach the cooling curves in FIG. 1, ideally to within a few degrees throughout the entire temperature range. However, because of the S-shaped form of the cooling curves and the large temperature range, such refrigeration processes are difficult to design. Pure component refrigerant processes, because of their flat vaporization curves, work best in the two-phase region. Multi-component refrigerant processes, on the other hand, have sloping vaporization curves and are more appropriate for the de-superheating and subcooling regions. Both types of processes, and hybrids of the two, have been developed for liquefying natural gas.
Cascaded, multilevel, pure component refrigeration cycles were initially used with refrigerants such as propylene, ethylene, methane, and nitrogen. With enough levels, such cycles can generate a net heating curve that approximates the cooling curves shown in FIG. 1. However, as the number of levels increases, additional compressor trains are required, which undesirably adds to the mechanical complexity. Further, such processes are thermodynamically inefficient because the pure component refrigerants vaporize at constant temperature instead of following the natural gas cooling curve, and the refrigeration valve irreversibly flashes the liquid into vapor. For these reasons, mixed refrigerant processes have become popular to reduce capital costs and energy consumption and to improve operability.
U.S. Pat. No. 5,746,066 to Manley describes a cascaded, multilevel, mixed refrigerant process for ethylene recovery, which eliminates the thermodynamic inefficiencies of the cascaded multilevel pure component process. This is because the refrigerants vaporize at rising temperatures following the gas cooling curve, and the liquid refrigerant is subcooled before flashing thus reducing thermodynamic irreversibility. Mechanical complexity is somewhat reduced because fewer refrigerant cycles are required compared to pure refrigerant processes. See, e.g., U.S. Pat. No. 4,525,185 to Newton; U.S. Pat. No. 4,545,795 to Liu et al.; U.S. Pat. No. 4,689,063 to Paradowski et al.; and U.S. Pat. No. 6,041,619 to Fischer et al.; and U.S. Patent Application Publication Nos. 2007/0227185 to Stone et al. and 2007/0283718 to Hulsey et al.
The cascaded, multilevel, mixed refrigerant process is among the most efficient known, but a simpler, more efficient process, which can be more easily operated, is desirable.
A single mixed refrigerant process, which requires only one compressor for refrigeration and which further reduces the mechanical complexity has been developed. See, e.g., U.S. Pat. No. 4,033,735 to Swenson. However, for primarily two reasons, this process consumes somewhat more power than the cascaded, multilevel, mixed refrigerant processes discussed above.
First, it is difficult, if not impossible, to find a single mixed refrigerant composition that generates a net heating curve that closely approximates the typical natural gas cooling curve. Such a refrigerant requires a range of relatively high and low boiling components, whose boiling temperatures are thermodynamically constrained by the phase equilibrium. Higher boiling components are further limited in order to avoid their freezing out at low temperatures. The undesirable result is that relatively large temperature differences necessarily occur at several points in the cooling process, which is inefficient in the context of power consumption.
Second, in single mixed refrigerant processes, all of the refrigerant components are carried to the lowest temperature even though the higher boiling components provide refrigeration only at the warmer end of the process. The undesirable result is that energy must be expended to cool and reheat those components that are “inert” at the lower temperatures. This is not the case with either the cascaded, multilevel, pure component refrigeration process or the cascaded, multilevel, mixed refrigerant process.
To mitigate this second inefficiency and also address the first, numerous solutions have been developed that separate a heavier fraction from a single mixed refrigerant, use the heavier fraction at the higher temperature levels of refrigeration, and then recombine the heavier fraction with the lighter fraction for subsequent compression. See, e.g., U.S. Pat. No. 2,041,725 to Podbielniak; U.S. Pat. No. 3,364,685 to Perret; U.S. Pat. No. 4,057,972 to Sarsten; U.S. Pat. No. 4,274,849 to Garner et al.; U.S. Pat. No. 4,901,533 to Fan et al.; U.S. Pat. No. 5,644,931 to Ueno et al.; U.S. Pat. No. 5,813,250 to Ueno et al; U.S. Pat. No. 6,065,305 to Arman et al.; and U.S. Pat. No. 6,347,531 to Roberts et al.; and U.S. Patent Application Publication No. 2009/0205366 to Schmidt With careful design, these processes can improve energy efficiency even though the recombining of streams not at equilibrium is thermodynamically inefficient. This is because the light and heavy fractions are separated at high pressure and then recombined at low pressure so that they may be compressed together in a single compressor. Generally, when streams are separated at equilibrium, separately processed, and then recombined at non-equilibrium conditions, a thermodynamic loss occurs, which ultimately increases power consumption. Therefore the number of such separations should be minimized. All of these processes use simple vapor/liquid equilibrium at various places in the refrigeration process to separate a heavier fraction from a lighter one.
Simple one-stage vapor/liquid equilibrium separation, however, doesn't concentrate the fractions as much as using multiple equilibrium stages with reflux. Greater concentration allows greater precision in isolating a composition that provides refrigeration over a specific range of temperatures. This enhances the process ability to follow the typical gas cooling curves. U.S. Pat. No. 4,586,942 to Gauthier and U.S. Pat. No. 6,334,334 to Stockmann et al. (the latter marketed by Linde as the LIMUM® 3 process) describe how fractionation may be employed in the above ambient compressor train to further concentrate the separated fractions used for refrigeration in different temperature zones and thus improve the overall process thermodynamic efficiency. A second reason for concentrating the fractions and reducing their temperature range of vaporization is to ensure that they are completely vaporized when they leave the refrigerated part of the process. This fully utilizes the latent heat of the refrigerant and precludes the entrainment of liquids into downstream compressors. For this same reason heavy fraction liquids are normally re-injected into the lighter fraction of the refrigerant as part of the process. Fractionation of the heavy fractions reduces flashing upon re-injection and improves the mechanical distribution of the two phase fluids.
As illustrated by U.S. Patent Application Publication No. 2007/0227185 to Stone et al., it is known to remove partially vaporized refrigeration streams from the refrigerated portion of the process. Stone et al. does this for mechanical (and not thermodynamic) reasons and in the context of a cascaded, multilevel, mixed refrigerant process that requires two separate mixed refrigerants. The partially vaporized refrigeration streams are completely vaporized upon recombination with their previously separated vapor fractions immediately prior to compression.
Multi-stream, mixed refrigerant systems are known in which simple equilibrium separation of a heavy fraction was found to significantly improve the mixed refrigerant process efficiency if that heavy fraction isn't entirely vaporized as it leaves the primary heat exchanger. See, e.g., U.S. Patent Application Publication No. 2011/0226008 to Gushanas et al. Liquid refrigerant, if present at the compressor suction, must be separated beforehand and sometimes pumped to a higher pressure. When the liquid refrigerant is mixed with the vaporized lighter fraction of the refrigerant, the compressor suction gas is cooled, which further reduces the power required. Heavy components of the refrigerant are kept out of the cold end of the heat exchanger, which reduces the possibility of refrigerant freezing. Also, equilibrium separation of the heavy fraction during an intermediate stage reduces the load on the second or higher stage compressor(s), which improves process efficiency. Use of the heavy fraction in an independent pre-cool refrigeration loop can result in a near closure of the heating/cooling curves at the warm end of the heat exchanger, which results in more efficient refrigeration.
“Cold vapor” separation has been used to fractionate high pressure vapor into liquid and vapor streams. See, e.g., U.S. Pat. No. 6,334,334 to Stockmann et al., discussed above; “State of the Art LNG Technology in China”, Lange, M., 5th Asia LNG Summit, Oct. 14, 2010; “Cryogenic Mixed Refrigerant Processes”, International Cryogenics Monograph Series, Venkatarathnam, G., Springer, pp 199-205; and “Efficiency of Mid Scale LNG Processes Under Different Operating Conditions”, Bauer, H., Linde Engineering. In another process, marketed by Air Products as the AP-SMR™ LNG process, a “warm”, mixed refrigerant vapor is separated into cold mixed refrigerant liquid and vapor streams. See, e.g., “Innovations in Natural Gas Liquefaction Technology for Future LNG Plants and Floating LNG Facilities”, International Gas Union Research Conference 2011, Bukowski, J. et al. In these processes, the thus-separated cold liquid is used as the middle temperature refrigerant by itself and remains separate from the thus-separated cold vapor prior to joining a common return stream. The cold liquid and vapor streams, together with the rest of the returning refrigerants, are recombined via cascade and exit together from the bottom of the heat exchanger.
In the vapor separation systems discussed above, the warm temperature refrigeration used to partially condense the liquid in the cold vapor separator is produced by the liquid from the high-pressure accumulator. The present inventors have found that this requires higher pressure and less than ideal temperatures, both of which undesirably consume more power during operation.
Another process that uses cold vapor separation, albeit in a multi-stage, mixed refrigerant system, is described in GB Pat. No. 2,326,464 to Costain Oil. In this system, vapor from a separate reflux heat exchanger is partially condensed and separated into liquid and vapor streams. The thus-separated liquid and vapor streams are cooled and separately flashed before rejoining in a low-pressure return stream. Then, before exiting the main heat exchanger, the low-pressure return stream is combined with a subcooled and flashed liquid from the aforementioned reflux heat exchanger and then further combined with a subcooled and flashed liquid provided by a separation drum set between the compressor stages. In this system, the “cold vapor” separated liquid and the liquid from the aforementioned reflux heat exchanger are not combined prior to joining the low-pressure return stream. That is, they remain separate before independently joining up with the low-pressure return stream. As will be explained more fully below, the present inventors have found that power consumption can be significantly reduced by, inter alia, mixing a liquid obtained from a high-pressure accumulator with the cold vapor separated liquid prior to their joining a return stream.